The Request Line: An Audit of America’s Broken Data-Sharing Promises
Photo: researcher email inbox data request laptop archive documents, via cdn.asp.events
Federal agencies now require many grant recipients to attach a data availability statement to their papers. The language is familiar: data are available upon request, or data are openly accessible through a repository. On paper, that sentence is a promise. In practice, it is often a forwarding address with no house.
Between August and October 2026, Knowledge Barrier-Less Press tested that promise. We drew a sample of 60 articles published in the preceding eighteen months that acknowledged support from the National Institutes of Health, the National Science Foundation, the Centers for Disease Control and Prevention, or the Agency for Healthcare Research and Quality. Each paper carried a data availability statement. Eight graduate students and two science librarians sent requests using the contact information printed on the paper. They identified themselves accurately, asked for the data and code needed to reproduce the headline finding, and followed up three times over six weeks.
The results were not a scandal of villains. They were a scandal of defaults.
What We Asked For
Our request was deliberately modest. We did not ask for every raw file, every interview transcript, or every protected patient record. We asked for the analytic dataset and the code that produced the paper’s central claim. Where privacy law applied, we asked for a description of the access process and an estimate of how long it would take. Where a repository was named, we asked for a direct link and a persistent identifier.
That is the minimum a serious data availability statement should deliver. A reader should be able to move from the claim in the abstract to the evidence behind it without hiring a lawyer or guessing which co-author still checks an old email address.
The Paper Trail
Of the 60 papers, 22 produced no substantive response after three attempts. Eleven responded with a promise but never sent data. Nine sent files that were incomplete, undocumented, or unusable without a codebook. Six required a data use agreement that was not mentioned in the paper and that imposed conditions beyond what the original consent process described. Four pointed to a repository landing page with no dataset attached. Eight delivered usable data and code within the six-week window.
Eight out of sixty is not a failure of individual generosity. It is a failure of design. The current system treats a data availability statement as a procedural box to check at submission. Journals rarely verify the claim. Funders rarely audit it. Universities rarely reward it. Researchers face no meaningful consequence when the statement turns out to be aspirational.
The missing responses also carried a cost that is easy to miss. Our student auditors spent hours drafting polite follow-ups, tracking bounced emails, and reading institutional data policies. That is unpaid labor extracted from people who are trying to learn how science works. For a journalist on a deadline, a community organizer preparing testimony, or a patient advocate comparing treatments, the same friction becomes a closed door.
The Cost of a Polite No
The most common phrase in our inbox was not no. It was soon. Soon became after the semester. After the semester became we are still cleaning the files. In several cases, the corresponding author had moved institutions, and the data lived on a laptop or drive that no one else could locate. In others, the request was routed to an office that did not know the paper existed.
These are mundane failures, but they compound into a public access problem. Taxpayers fund the study, the training, the infrastructure, and the publication. Then they are asked to accept a sentence as proof that the evidence is within reach. A statement that cannot be tested is not transparency. It is branding.
Privacy is a real constraint, and it should not be dismissed. Health data, school records, and community interviews require careful handling. But privacy is not an excuse for an unlisted process. A serious statement can say which data are restricted, why, who decides access, how long the review takes, and what a requester can do if the answer is no. That is not a burden on science. It is the infrastructure of credible science.
Why the Honor System Fails
The deeper problem is that data sharing is still treated as a virtue rather than a condition of funding. Agencies issue guidance. Journals publish policies. Institutions hold workshops. Then everyone returns to a reward system that counts publications, citations, and grant dollars, not whether a stranger could reproduce the finding.
That is why enforceable standards matter more than better intentions. A machine-readable data availability statement would let funders and journals check whether a claimed repository actually contains the promised files. A public dashboard could show response rates and unresolved requests by institution and funder. A funding condition could require a named data steward after the corresponding author leaves. A safe harbor could protect researchers who share de-identified data in good faith from liability when they follow approved protocols.
None of these ideas is radical. They are the ordinary plumbing of a public system. The radical move would be to keep pretending the plumbing works because the sentence sounds right.
What Enforceable Access Looks Like
The fix should begin where the money begins. Federal agencies should require data availability statements to include a persistent identifier, a named contact, a plain-language access process, and a timeline for response. Journals should reject statements that cannot be verified. Universities should count reproducible data stewardship in promotion and tenure, not just in grant applications. Funders should publish annual compliance data so the public can see which institutions keep their promises.
There is also a role for Congress. When lawmakers appropriate billions for biomedical and behavioral research, they can ask a simple question: can the public find the evidence? A public ledger of data requests, responses, and denials would answer that question without exposing sensitive records. It would also shift the burden from the requester to the institution that accepted the public money.
The open access movement has spent years arguing about who can read the paper. The next fight is about who can inspect the evidence. A paper without usable data is a brochure. A data availability statement without a working request line is a slogan. Taxpayers deserve better than a polite no dressed up as openness.